ABSTRACT The Finite Element Method (FEM) is a powerful tool for simulating non‐linear magnetic devices, but its high computational cost becomes a significant limitation when multiple simulations are required, such as in design optimisation or real‐time control. To address this challenge, this paper proposes a non‐intrusive surrogate modelling framework designed to significantly reduce computation time with minimal accuracy loss. The approach is based on a low‐dimensional parameterisation of the solution space using Proper Orthogonal Decomposition (POD), combined with machine learning‐based interpolation in the reduced space. The approach is validated by creating a surrogate model of a nonlinear inductor near an iron piece, reconstructing the field distribution as a function of current, frequency, and relative position between the two. Two surrogate modelling techniques, Gaussian Process Regression (GPR) and Feedforward Neural Networks (FNN), are investigated and compared, with particular attention to their performance under data‐scarce conditions, common in engineering workflows due to the high cost of generating training data. Numerical results demonstrate that GPR provides more accurate approximations than FNN, especially when few FEM simulations are available. The findings highlight the potential of POD and GPR as efficient and reliable tools for accelerating the simulation and optimisation of magnetic devices.

Electric Vehicles (EVs) must allocate a substantial portion of their battery energy to maintain thermal comfort in the passenger cabin, especially in cold ambient conditions where conventional heating systems can significantly reduce driving range. This paper presents an embedded smart actuator architecture for localized thermal management in EV cabins. The proposed system integrates a microcontroller-based control board directly with an etched foil heating element, transforming a passive heater into a controlled node capable of autonomous power regulation. The control law exploits the Predicted Mean Vote (PMV) index as a global comfort metric, computed from real-time microclimatic measurements acquired through a dedicated sensing station. A dual-loop cascade control architecture is implemented: an outer controller translates the PMV error into a temperature setpoint for the heater, while an inner adaptive controller regulates the heater surface temperature through PWM-driven power modulation. The system was experimentally validated inside a climatic chamber hosting an electric vehicle under cold conditions. Preliminary results demonstrate stable temperature regulation without overshoot and show the control architecture’s ability to dynamically respond to variations in the cabin thermal comfort index. Although the validation focuses on a single heating element, the proposed architecture represents a scalable framework for distributed networks of smart heating nodes, enabling energy-efficient localized heating strategies for next-generation electric EV cabins.

Thermal comfort inside vehicle cabins directly impacts driver well-being, attention, and overall road safety. Modern Heating Ventilation and Air Conditioning (HVAC) systems increasingly rely on quantitative comfort models like the Predicted Mean Vote (PMV), which requires both environmental and personal parameters. While environmental variables can be measured using standard sensors, personal parameters, especially clothing insulation, are typically approximated with static seasonal values, leading to significant uncertainty in the comfort estimation process. This paper presents a vision-based system for estimating driver clothing insulation in automotive cabins using a low-cost camera and deep learning models. The approach involves a YOLOv8 Nano detector to identify clothing regions in the camera frame and two MobileNetV3 classifiers to infer garment categories and fabric types. The predicted garments are then mapped to clothing insulation values using standardised reference tables, resulting in an estimate of the driver’s total clothing insulation level. The system was trained on subsets of the DeepFashion and DeepFashion2 datasets and evaluated on benchmark data and in real in-vehicle acquisition sessions using a webcam installed inside an electric vehicle cabin. The garment classifier achieved satisfactory classification accuracy on the benchmark dataset, while the end-to-end pipeline obtained a clothing insulation estimation error of 0.186clo (0.160clo excluding two outlier sessions) during in-cabin experiments. These results demonstrate the feasibility of estimating clothing insulation using computer vision in constrained automotive environments, paving the way for occupant-aware HVAC systems capable of real-time PMV-based thermal comfort control.

Induction heating is widely used in industrial processes such as forging, hardening, and additive manufacturing, but its accurate numerical simulation requires coupled electromagnetic and thermal finite element analyses with nonlinear, temperature-dependent material properties. This work proposes a deep learning surrogate model based on a convolu-tional neural network for TEAM Workshop Problem 36, a reference benchmark for nonlinear magneto-thermal induction heating. A database of more than 40,000 finite element solutions was generated by varying the supply current from 2 to 6 kA and the frequency from 2 to 6 kHz, while accounting for transient nonlinear effects, including the Curie transition. The network, composed of 24 layers with transposed convolutions, batch normalization, and dropout, maps current, frequency, and time to radial temperature distributions in the steel billet. For most operating conditions, the model achieves mean absolute percentage errors of about 6–7% for radial line in the middle of the billet and about 10% for radial line close to the billet end. Larger discrepancies occur during the early heating stage and near the Curie temperature. Prediction times are reduced by three to four orders of magnitude with respect to a single finite element analysis. The results indicate that the proposed surrogate enables fast temperature estimation for optimization, digital twins, and closed-loop control of induction heating systems.

Electroporation outcomes are governed by the local electric field distribution and transmembrane voltage, both of which may be altered by nanoscale elements positioned near the cell membrane. In this study, we developed a two-dimensional finite-element electromagnetic model to investigate the effect of a membrane-proximal silica-coated superparamagnetic iron oxide nanoparticle cluster during a trapezoidal electroporation pulse. The model couples electric and magnetic field components with a membrane electroporation formulation based on Smoluchowski-type pore-density dynamics. Simulations were performed with and without a nanoparticle positioned 5 nm from the membrane, considering different cytosol and extracellular medium conductivities. The results show that the nanoparticle induces a highly localized perturbation of the electric field, whose magnitude depends on the sampling region and conductivity contrast. Transmembrane voltage is modestly and transiently modulated during pulse rise time, whereas the effect is limited during the pulse plateau. Pore-density analysis further indicates that the nanoparticle does not induce a generalized increase in electroporation-related parameters and may locally reduce pore density near the nanoparticle–membrane interface. Overall, the model identifies transient and conductivity-dependent nanoscale field redistribution caused by membrane-proximal silica-coated magnetic nanoparticles, while highlighting the need for three-dimensional modeling and experimental validation before inferring electroporation enhancement.

Industrial production still relies heavily on thermal processes that predominantly use fossil fuels for energy. This has significant consequences for primary energy use and greenhouse gas emissions. Meanwhile, rapid advances in electrotechnologies—defined as processes that use electrical energy to transform materials through internal heat dissipation (inductive, conductive, or dielectric/microwave) or heat transfer via resistance and infrared systems—are paving the way for a transition to a non-fossil fuel-based energy supply across a wide range of temperatures and power densities. However, replacing fuel with electricity is not simply a case of making a straightforward substitution; the feasibility of this change is determined by process requirements, constraints on installation space and grid connection, the reliability and volatility of the electricity supply, and economics. This paper therefore proposes a simple, decision-oriented methodology to assess the feasibility of defossilisation from energetic and economic perspectives. The methodology centres on a “substitution coefficient” that compares the amount of fossil energy substituted by a given amount of electrical energy and benchmarks this against the primary energy intensity of electricity generation. The methodology is demonstrated using case studies from energy-intensive sectors such as cement production (using resistance and microwave methods), steel strip processing (with inductive boosting combined with resistive holding) and metal melting for cast iron and aluminium. The case studies show under which conditions electrification can be implemented as a drop-in substitute, a hybrid booster or an enabler of new production models. The results indicate where electrotechnologies can deliver primary energy savings and CO2 reductions today and outline the conditions under which their advantages will increase as power systems become more decarbonised.

A cascade deep learning approach is proposed for optimizing the design and control of a dual-frequency induction heating system used in semiconductor manufacturing. The system is composed of two independent power inductors, fed at different frequencies, to achieve a homogeneous temperature profile along a graphite susceptor surface, crucial for enhancing layer quality and integrity. The optimization process considers both electrical (current magnitudes and frequencies) and geometrical parameters of the coils, which influence the power penetration and subsequent temperature distribution within the graphite disk. A two-step procedure based on deep neural networks (DNNs) is employed. The first step, namely optimal design, identifies the optimal operating frequencies and geometrical parameters of the two coils. The second step, namely optimal control, determines the optimal current magnitudes. The DNNs are trained using a database generated through finite element (FE) analysis. This deep learning-based cascade approach reduces computational time and multiphysics simulations compared to classical methods by reducing the dimensionality of parameter mapping. Therefore, the proposed method proves to be effective in solving high-dimensional multiphysics inverse problems. From the application point of view, achieving thermal uniformity (±7% fluctuation at 1100 °C) improves layer quality, increases efficiency, and reduces operating costs of epitaxy reactors.

With the increasing volumes of spent lithium-ion batteries from electric vehicles and the concurrent increase in raw materials cost for cathode production, finding effective methods for recycling battery materials has become critically important. This study investigated a pyrometallurgical approach using microwave irradiation to achieve carbothermal reduction of LiCoO2. FactSage thermodynamic calculations were performed for process simulation and an infrared thermal camera was employed for temperature measurements, allowing the authors to optimize the process parameters to obtain metallic cobalt. Specifically, the research included microwave experiments on mixed black mass samples of anode and cathode materials in different proportions, treated at varying power levels and exposure times under air atmosphere. The effect of the process parameters and therefore of the temperature on microstructure was studied with SEM-EDS and XRD analysis. The feasibility of a wet magnetic separation method between cobalt and lithium compounds formed during the reaction was also evaluated. The results obtained from the final separation process indicated that individual compounds can be obtained at the end of the cycle; moreover, the optimization of time, temperature, and graphite additions during the tests allowed the authors to obtain promising results.

In the paper, finite element analyses are used to study the effect of media inhomogeneity on the electric field distribution in a sample composed of extracellular matrix and cells, including cells with a lipid droplet in the cytoplasm. In the model, voltage pulses, typical of electrochemotherapy treatments, are considered. The electric field distribution around cells is computed numerically; moreover, the transmembrane potential at the cell membrane is evaluated. In order to study cell behavior in different conditions either the presence of the collagen or the lipid droplet inside the cell cytoplasm are considered. The conductivity of the extracellular medium varied between plain collagen to myxoid, i.e., a gel-like matrix; the combinations of the two cases (a collagen shell around cells immersed in a myxoid environment) is considered too. An increase in the transmembrane potential was found in the case of higher aggregate and collagen presence. However, it is already known that, increasing the cell aggregation, the transmembrane potential increases as well as in the case of the presence of the collagen shell. In the paper, the effect of a lipid droplet is studied too.

In this work, silicon recovered from a PV panel recycling facility was utilized as a reactant for the synthesis of ferrosilicon alloys via microwave treatment in the presence of iron oxides and graphite. Microwave-driven processing was further explored using ladle furnace slag from secondary steelmaking as a reactive component. Different amounts of reactants, time and microwave power were studied. Microwave heating allowed to reach temperatures higher than 900 degrees C after few minutes of treatment at a power of 600 W. The materials were characterized by SEM-EDS and XRD. The proposed process allowed to obtain a ferrosilicon alloy with an amount of 21% Si. This study explored the use of three different types of waste, PV silicon, iron oxide and ladle furnace slag to obtain a ferroalloy FeSi which can be afterwards employed in the steelmaking process.

Induction heating processes involve complex multiphysical interactions between electromagnetic and thermal models, often characterized by nonlinear materials. The finite element method is a common approach to tackling such problems, but its computational complexity may become a burden for optimization loops or real-time monitoring. This work proposes a surrogate modeling approach that combines proper orthogonal decomposition and Gaussian process regression, suited for nonlinear and temperature-dependent magnetic materials. The approach is applied to reduce the complexity of the electromagnetic model of the testing electromagnetic analysis method (TEAM) problem 36, consisting of a copper coil heating through induction a steel billet. Results show how the non-intrusive machine learning approach can accurately reconstruct the field distribution, offering a viable first step towards fast multiphysical simulations involving nonlinear magnetic materials.

Uniform electric field distribution in electroporation is crucial for achieving the effective localized delivery of drug molecules. Currently, in vitro electroporation studies on adherent cells lack a standardized test setup for obtaining consistent and repeatable results, unlike in vitro electroporation studies on cell suspensions that use electroporation cuvettes, which provide uniform electric field distribution. Considering this, we designed, built, and tested a novel inset design for low-volume round well plates, such as the 24- and 96-well plates which are most commonly used in cell culture labs. The inset design was realized using 3D printing and experimentally tested using potato phantoms and HeLa cells. Finite element analysis (FEA) was used to compute the electric field distribution in the round well plates with and without the inset. The FEA indicated that the electric field contour map at the bottom of the well with the inset had a more uniform electric field distribution, with an average value close to the expected 1000 V/cm. In contrast, it was only 840 V/cm without the inset, indicating non-uniform electric field distribution. Uniform electric fields were also obtained using the inset for the potato phantoms and the HeLa cells, indicating the merit of the inset and its usability with low-volume cell culture well plates, which enable the transfer of cells for various assays without additional steps, as well as its cost-effectiveness.

This article presents the application of proper orthogonal decomposition (POD) and Gaussian process regression (GPR) to develop a fast and accurate macromodel for predicting electromagnetic fields and forces in an electromagnetically levitated aluminum billet. The finite element method (FEM) was used to create a 2-D model of the device, extracting the current density and magnetic field distributions in the billet for different positions and frequencies. POD was applied to reduce the dimensionality of the FEM data, while GPR was employed to predict the reduced-order model coefficients for new input parameters. The resulting surrogate model significantly reduces computation time from 8 min to 52 ms, while maintaining a high level of accuracy, providing full-field predictions of the quantities of interest. The model was validated for both field and force predictions, demonstrating its potential to accelerate device study and optimization, while paving the way toward its application as a digital twin of the device.

In this paper, a multi-physics case study belonging to the class of induction heating problem is considered. Finite Element Analysis is used to evaluate the temperature along a line on a graphite disk heated by two power inductors. In order to build a surrogate field model of the device, i.e., to compute the temperature profile on the disk, given the amplitudes and frequencies of the supply currents, three methods have been used (Support Vector Regression (SVR), fully connected Neural Network (NN) and Gaussian Process Regression (GPR)). In turn, to solve the inverse problem, i.e., to identify frequencies and currents of the two coils, given a prescribed temperature profile, two approaches have been implemented. The former is an optimization approach based on a multi-objective formulation, solved by means of the NSGA-II algorithm; the latter is a two-step procedure, based on fully connected Deep Neural Networks (DNNs), solving an optimal design problem first and, subsequently, an optimal control problem.

In the paper an induction heating problem is considered. The Finite Element analysis is used to evaluate the temperature along a line on a graphite disk heated by a power inductor. Subsequently, a fully-connected Neural Network (NN) is trained with the aim of implementing a surrogate model able to compute the temperature profile on the disk, given the amplitude and frequency of the supply current.

Purpose Recent progress in additive manufacturing methods alleviated manufacturing constraints on devices. Topology optimization (TO) methods can leverage these reduced limitations and this paper aims to study the use of these algorithms in induction heating for injection molding. Design/methodology/approach In this paper, TO is used to reduce the volume of ferrite in an injection molding tool while aiming at maximizing the performance of the device. Characteristics of the proposed solution such as efficiency and power density are compared to the ones of the original device. Findings The study shows that it is possible to reduce significantly the amount of ferrite used without impacting the efficiency. The thermal performances of the proposed solution present also slight improvements compared to the original solution. Originality/value Optimization algorithms are important for understanding how to design efficient electrical devices. In this paper, the application of TO for injection molding applications presents a new perspective in designing such components.

Purpose The purpose of the study is to design the compensation network of a dynamic wireless power transfer system, considering the movement of the receiving coil along an electrified track with a large number of inductors buried on the road. Design/methodology/approach A finite element model has been developed to calculate the self-inductances of transmitting and receiving coils as well as the mutual inductances between the receiving coil and the transmitting ones in the nearby and for various relative positions. The calculated lumped parameters, self-inductances and mutual inductances depending on the relative positions between the coils, have been considered to design the compensation network of the active coils, which is composed of three capacitive or inductive reactances connected in the T form. The optimal values of the six reactances, three for the transmitting coils and three for the receiving one, have been calculated by resorting to the Genetic Algorithm NSGA-II. Findings In this paper, the results obtained by means of the optimizations have broadly discussed. The optimal values of the reactances of the compensation networks show a clear trend in the receiving part of the circuit. On the other hand, the problem seems very sensitive to the values of the reactances in the transmitting circuit. Originality/value Dynamic wireless power transfer system is one of the newest ways of recharging electric vehicles. Hence, the design of compensation networks for this kind of systems is a new topic, and there is the need to investigate possible solutions to obtain a good performance of the recharging system.